OpenAI rolls out ChatGPT Agent to handle complex computer tasks
The agent mode is available starting Friday to Pro, Plus, and Team subscribers through a tools dropdown in the chat interface.
OpenAI has rolled out ChatGPT Agent, its latest product that enables users to assign multi-step tasks to the AI chatbot across various applications and websites.
The newly launched feature will allow users to ask the AI chatbot to make personalised requests. These include accessing calendars and briefing on meetings based on news, planning and buying ingredients to make a Japanese breakfast for four, or analysing competitors and creating presentations, the company said.
ChatGPT Agent can browse websites, filter search results, prompt users to log in securely when needed, run code, conduct analysis, and deliver editable slideshows and spreadsheets summarising its findings.
“Most importantly, you’re always in control. ChatGPT requests permission before taking actions of consequence, and you can easily interrupt, take over the browser, or stop tasks at any point.While ChatGPT agent is already a powerful tool for handling complex tasks, today’s launch is just the beginning. We’ll continue to iteratively add significant improvements regularly, making it more capable and useful to more people over time,” reads the company’s blog.
The new capability combines three separate AI tools — Operator's website interaction abilities, Deep Research's information synthesis skills, and ChatGPT's conversational intelligence.
It acts as a unified system that operates through a virtual computer environment, allowing it to shift between reasoning and action to complete workflows.
The agent mode is available starting Friday to Pro, Plus, and Team subscribers through a tools dropdown in the chat interface. Users can activate the feature at any point during conversations.
The Sam Altman-led firm addressed limitations of OpenAI's previous tools, where Operator excelled at web interactions but struggled with deep analysis. In contrast, deep research could synthesize information but couldn't interact with websites or access authenticated content.
The launch comes as competition heats up, with tech giants racing to build AI systems capable of performing multi-step and challenging tasks independently across digital platforms.
Last week, AI search startup Perplexity rolled out Comet, its AI-powered web browser, making a big move to compete against Google’s longstanding dominance in the web browser space.
Similar to ChatGPT Agent, Comet lets users ask queries “out loud” and is capable of navigating websites, comparing information, and executing tasks on their behalf.